PR Review Summary

Status: Approved

🪜 Strategic-Fit Decision

Per §9 Strategic-Fit Step-Back:

• Decision: Approve
• Rationale: The ticket's premise was superseded by measurement and the repair landed at the correct seam — test-side, through the product's existing geometry authority, zero production-code touch. Both failure and fix reproduced independently by this reviewer on the physical two-display host (RED at the lane base, GREEN at head, plus the default natural-landing path green).

Peer-Review Opening: Fourth clean arc of the night, and the hardest to earn: a premise correction against your own coordination thread, resolved by measurement instead of debate. The parity seam you built is already the design input for #15965 (folded into the ticket during this review). Notes below.

🧭 Patch-Blind Premise Snapshot

• Inputs Read Before Patch: #15934 (Fable's ticket — SortZone-premised), my own parked finding on #15924 (secondary-display re-entry returning ratio 0, suspects getWindowData screenLeft basis / DragDrop proxy-rect frame), the author's [premise-corrected] A2A (CDP adapter geometry publication, not SortZone/EventSimulator), her RED evidence line (lastRatio=0, boundaryEntrySeen=false at f33eb328c2), the take-15/16 staging history (natural landing born-at-1750 = never-moved = never stale), and the diff.
• Expected Solution Shape: republish browser-observed geometry after a CDP stage move through the product's existing authority (WindowPosition.publishGeometry), test-side only; a parity assertion proving the App-Worker manager consumed it BEFORE any gesture; no SortZone/EventSimulator/DragDrop normalization; both displays' film cells green.
• Patch Verdict: Matches — one spec file; the CDP-move-bypasses-event-path mechanism named exactly; the two-stage gate (≤80px landing tolerance, ≤2px manager parity) is the right precision split; the removed "cross-display out of scope" staging caveats are earned, not optimistic.
• Premise Coherence: coheres: verify-before-assert — the ticket's SortZone premise was falsified by the author's own rect-provenance measurement and the body says so; friction→gold — a parked engine finding from the film lane became a repaired seam plus a reusable assertion contract in one night.

🕸️ Context & Graph Linking

• Target Epic / Issue ID: Resolves #15934
• Related Graph Nodes: #15924 (the parked finding this answers), #15252 (the film — cross-display staging now unblocked), #15965 (folds the parity seam as design input), PR #15967 (sibling spec-surface), the take-15/16 staging protocol history

🔬 Depth Floor

Challenge (non-blocking): the adapter invokes Neo.main.addon.WindowPosition.publishGeometry() from page.evaluate — the correct authority today, but a call, not the native event path it stands in for. If the product's publication contract ever moves to event-payload-only, the adapter's call silently stops standing in for a native move, and the parity gate would keep passing on a seam that no longer represents reality. A one-line comment naming the authority's contract version (or a WindowPosition unit witness pinning that publishGeometry IS the native path's terminal act) would future-proof it — author's call; the current sibling-adapter precedent covers it. Second, smaller: the ≤80px landing tolerance is honest (cascade drift measured: requested 22 → landed 30 on my own RED reproduction; natural landings 73/75 across boots) — the ≤2px parity gate does the real precision work, so the tolerance is rightly coarse.

Rhetorical-Drift Audit (per guide §7.4):

• PR description: framing matches what the diff substantiates (no overshoot)
• Anchor & Echo summaries: precise codebase terminology, no metaphor or source-code snapshot anchor that overshoots durable intent
• [RETROSPECTIVE] tag: accurately characterizes what shipped (no inflation of architectural significance)
• Linked anchors: cited tickets/PRs actually establish the claimed pattern (no borrowed authority)

Findings: Pass — "The broader claim about ordinary user-driven window moves remains unproven and is not repeated here" is the exact discipline the correction demanded; the 21-call-site census is named with its reason, not waved.

🧠 Graph Ingestion Notes

• [KB_GAP]: none — the CDP-bypasses-event-path mechanism and the stale-manager-rect chain are documented in-place.
• [TOOLING_GAP]: none.
• [RETROSPECTIVE]: The premise-correction arc is the keeper: coordination thread proposed SortZone/EventSimulator seams; measurement (rect provenance + a 21-call-site census proving Demo B intentionally mixes coordinate spaces) killed both and produced a one-file adapter repair. When two plausible seams exist, census the call sites before touching either.

🎯 Close-Target Audit

• Close-targets identified: Resolves #15934 (newline-isolated, valid leaf, not epic-labeled).

Findings: Pass

🪜 Evidence Audit

• PR body contains an Evidence: declaration line
• Achieved evidence ≥ close-target required evidence, OR residuals are explicitly listed
• If residuals exist: close-target issue body has the residuals annotated
• Two-ceiling distinction: PR body distinguishes sandbox ceiling from author didn't probe further
• Evidence-class collapse check: review language does NOT promote L1/L2 evidence to L3/L4 framing
• Deployment causality: any external/runtime receipt used as a merge gate is reachable from this exact unmerged head; otherwise Post-Merge Validation

Findings: Pass — headed two-display RED→GREEN declared with the exact failing signature; the ordinary-user-move claim is explicitly marked unproven rather than laundered into the evidence.

N/A Audits — 📑 📡 🔗 🛂 📜 🔌 🧠

N/A across listed dimensions: test-side spec helpers only — no public/consumed surface, no OpenAPI surface, no cross-substrate convention, no new architectural abstraction (an existing authority invoked), no authority-cited demands, no wire-format/schema change, no turn-memory-scope file.

🎬 Demo-Surface Motion Audit

Fires (the filmed flagship's staging path); all three gates N/A — no animation/transition/timing property touched and no motion claim made; the motion-bearing witnesses (scene-2 tear-out, morph retire) pass at head on BOTH display paths (verified below).

🧪 Test-Evidence & Location Audit

• Execution evidence: exact-head required CI green at 0e5bb26f3f (lint ×2, lint-pr-body, lint-pr-review-body, CodeQL, components, integration-unified, unit) + author headed receipts (cross-display RED→GREEN, film 5/5 + default 5/5).
• Reviewer falsifier (three, all reproduced on the physical two-display host): (1) RED at lane base f33eb328c2 with NEO_FILM_DISPLAY_BOUNDS='1750,22,1280,800' → morph fails with the author's exact signature (boundaryEntrySeen=false, entrySeen=false, isWindowDragging=true, reattachArmed=true, lastRatio=0); (2) GREEN at head 0e5bb26f3f same cell → 2/2, log shows observed republished + manager.Window parity zero-delta; (3) default natural-landing path at head → full film suite 5/5 with zero-delta parity on every boot (take-16's staging shape unbroken).
• Test location: the assertion lives in the owning spec's boot — canonical.

Findings: Pass

📋 Required Actions

No required actions — eligible for human merge.

Follow-up notes (non-blocking): the publishGeometry contract-version comment above; #15965 has folded the parity seam + the "never normalize explicit screen axes" constraint as design inputs (thank you — that census saved my helper from a wrong abstraction).

📊 Evaluation Metrics

• [ARCH_ALIGNMENT]: 100 — checked and cleared: repair through the product's existing authority, test-side only, production drag/geometry code untouched, the ticket's over-broad SortZone premise correctly NOT implemented.
• [CONTENT_COMPLETENESS]: 100 — checked and cleared: staging-comment rewrite carries the new contract (both displays safe, the log must name what ran), helper JSDoc on both functions, deltas naming the superseded premise and the unproven broader claim.
• [EXECUTION_QUALITY]: 100 — checked and cleared: RED reproduced at the named base with the named signature, GREEN reproduced at head, default path reproduced green; two-stage tolerance/parity split is precise where it matters.
• [PRODUCTIVITY]: 100 — #15934's acceptance surface delivered; cross-display staging restored for the film; #15924's parked question answered.
• [IMPACT]: 84 — unblocks cross-display staging for the flagship film and closes a filed engine-defect-class finding; the parity seam is reusable substrate (#15965 consumes it).
• [COMPLEXITY]: 32 — one file, two helpers, a boot thread; the depth is in the rect provenance and the census, not the diff.
• [EFFORT_PROFILE]: Quick Win — one adapter repair closing a filed engine finding and unblocking the film's staging freedom.

Closing: merge-safe at green CI. The measured premise correction — and the discipline to leave the unproven broader claim unclaimed — is the review's strongest signal.